A High Precision Iris Feature Extraction and Its Application in Iris Recognition * ** 84 E-mail: *cld3@giga.net.tw, **pierre@isu.edu.tw ( ) ( ) CASIA ABSTRACT In this paper, a novel technique is proposed for high performance iris feature extraction. First, we elaborate a simple and fast iris location method and extract an iris image by the center coordinate of pupil. The iris image will be stretched into a rectangle block and the block is segmented many parts. We adopt Sobel transform to enhance iris texture, vertical projection to obtain one dimension energy signal, and one dimension wavelet transform to reduce the 4 feature vectors of the energy signal and restrain the 8 4 high frequency noise. Finally, probabilistic neural 7 8 network (PNN) is regarded as a classifier for iris recognition. A comparative experiment of existing methods for iris recognition is evaluated on CASIA iris image databases. The experimental results reveal the proposed algorithm provides superior performance in iris recognition, but it has still small 993 Daugman feature dimension and recognition efficiency. These [-3] prove the proposed method is suitable for embedded (-D Gabor wavelet transform) system or real-time system in resource-constrained. Keywords : iris recognition, wavelet transform, probabilistic neural network 48bits Daugman.
[4-5,4,6-7] (elaborate) (a) CASIA. (a) (pupil) (sclera) (b) (c). (a) (b) (c) [4]. LL /4. (b) T 3. T = w p () 3. (Circular iris block image) P w B A (a), if A( > T B( = (), if A( < T (b) E ( ( = B( i + i j + j E (3) ii = jj = B(, if E( > 4 B ( = (4), otherwise ( E >4 ( B 4 ( i B, (c) 3. (d) (c) (e). (a) (b) (c) (d) (e) (b) (a)
( (c)) ( (d)) (normalize)( (e)) (Multi-resolution analysis, MRA) 3 s H L cd (Detail Signal), ca (Approximate Signal) 3. (blur) cd [ n] = s[ k ] H[ n k ] D = s, H (9) k= 3 [3] ca[ n] = s[ k] L[ n k] A = s, L () k = (5) 4. (a) 3.3 (b) 3. (a) (b) 3.3 3 mxn 4. g xn L g (Neural-based Classifier for Iris xn G = M O M (6) Recognition) 4. g mx L g mxn D.F.specht 988 (Probabilistic Neural Network, PNN)[8] 5 input units pattern units (summation s units) output units s = m [-5] i g ij, < i < n, < j < m (7) j = S S = [ s L s n ] (8). m n n. (concentrate) 3. [4] σ 3.4 4 (Wavelet Transform, WT) [7,9] σ 4
N t ( xi y Ai) ( xi y Ai) () P f A( X) = ( ) p / p exp( ) (π ) σ N i = σ N PNN f A (X ) : X A N P p : σ : P N : A (6) X : y Ai : A i f A f A X P 4.4 (Matching) (verify) PNN PNN PNN P P P fa (Contingence Table) H H ( ) ( ) fb (False Rejection Rate, FRR) H (False Acceptance Rate, 5. FAR) CHEN CHU [4] (threshold) 4. FAR = / FAR = / FAR FRR 5 P FAR FRR (Equal Error 4.3 (Identification) Rate, ERR) (recognition rate) [-5] 5. (Particle Swarm Optimization, PSO)[8] X Y X PNN
PSO PSO. (%) 99.3 PSO (%) 85 PSO % 99.3% PSO 85 PSO (Social-Only model) 6. (Matching) (Cognition-Only model): (One-class ) Social-Only Model: recognition) 3 Vid = Vid + c rand( ) ( Pgd X id ) () ) Cognition-Only Model: FAR Vid = Vid + c rand( ) ( Pid X id ) (3) FRR. (FAR/FRR) 3) PSO Combine Model: (%) Vid = Vid + c rand( ) ( Pid X id ) FAR(%) FRR(%) 35.8. + c * rand( ) ( P gd X id ) 36.9..3 X id = Xid + V (4) id 37...69 d i 4..9 V i X i EER 4 P i P g 3. (EER) c c EER (%).54 EER (%). (ms) < 6. FAR FRR35.8%4%FAR.%.% FRR.%.9% Mayank Valsa CASIA [] 3 EER=.% Mayank Valsa 43 CASIA [] 8 8 ms 7 3 6.3 3x8 AMD.G Hz(.8G)5M ddram Windows Mayank Valsa[] XP Matlab 6.5 CASIA (iris coding) 6. (Identification) Mayank Valsa 3 4 4.
Multiple Features for High Performance Face FAR/FRR(%) (%) Recognition System ", 4 International Computer Symposium (ICS4), Taipe Dec Avila [7].3/.8 97.89 4. Li Ma [4]./.98 98 6 Christel-Loic Tisse, Lionel Torres,Michel Robert, Tissue [6].84/8.79 89.37 Person Identification Based on Iris Patterns, Daugman []./.9 99.9 Proceedings of the 5 th International Proposed./.69 99.3 Conference on Vision Interface,. 4 7 C. Sidney Burrus Ramesh A.Gopinath Hai ta Guo, Introduction to Wavelets and Wavelet Daugman Transforms: A Primer, 998. 8 D.F.Specht, Probabilistic Neural Network for Classification, Map, or Associative Memory, Daugman Proceeding of the IEEE International Conference on Neural Network, vol., pp55-53, 988. (iris coding) 9 Jaideva C.Goswami and Andrew K. Chan Fundamentals of Wavelets 999. J. Daugman, Biometric Personal Identification System Based on Iris Analysis, United States Patent, no. 5956, 994 J. Daugman, High Confidence Visual Recognition of Person by a Test of Statistical Independence, IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 5, no., pp48-5, Nov. 993 J. Daugman, Demodulation by Complex-Valued Wavelets for Stochastic Pattern Recognition, Int l J, Wavelets, Multiresolution and Information Processing, vol., no., pp. -7, 3. CASIA 3 J. Daugman, Statistical Richness of Visual % Phase Information: Update on Recognition Persons by Iris Patterns, Int l J. Computer EER.% Vision, vol. 45, no., pp. 5-38.P.W. Hallinan Recognizing Human Eyes, Geometric ms Methods Comput. Vision. Vol. 57, pp. 4-6, 99. 4 L. Ma, Tan, Tieniu, Wang, Yunhong, Iris CASIA iris database. Institute of Automation, Chinese Academy of Sciences, [Online].http://www.sinobiometrics.com/casiairis.htm. Chia -Te CHU and Ching-Han CHEN, The Application of Face Authentication System for Internet Security Using Object-Oriented Technology, Journal of Internet Technology. (Accepted) 3 Ching-Han CHEN, Chia -Te CHU, " An High Efficiency Feature Extraction Based on Wavelet Transform for Speaker Recognition ", 4 International Computer Symposium (ICS4), Taipe Dec 4. 4 Ching-Han CHEN, Chia-Te CHU, " High Efficiency Iris Feature Extraction Based on -D Wavelet Transform ", 5 Design automation and test in Europe (DATE5), Munich,Germany, March, 5. 5 Ching-Han CHEN, Chia -Te CHU, " Combining Recognition Using Circular Symmetric Filters, Processing of THE 6 Th International Conference on Pattern Recognition, vol., pp. 44-47,. 5 Mayank Vatsa, Richa Singh, and P.Gupta, Comparison of Iris Recognition Algorithms, Proceedings of ICISIP'4, India, pp.354-358, 4. 6 R.Wildes, J.Asmuth, G. Green, S. Hsu, R. Kolczynsk J. Matey, and S. McBride, A Machine-Vision System for Iris Recognition, Machine Vision and Application, vol. 9, pp. -8, 996. 7 Sanchez-Avila C., Sanchez-Reillo R.; de Martin-Roche D., Iris Recognition for Biometric Identification Using Dyadic Wavelet Transform Zero-Crossing, Proceedings of the IEEE 35 th International Carnahan Conference on Security Technology, pp. 7-77,. 8 J.Kennedy et al.,"particle Swarm Optimization", Proc of IEEE Int. Conf. Neural Networks, vol. IV, pp.94-948, 995.